Alkaline pH stress invokes a potent and fast transcriptional response in Saccharomyces cerevisiae that includes many genes repressed by glucose. Certain mutants in the glucose-sensing and -response pathways, such as those lacking the Snf1 kinase, are sensitive to alkalinization. In the present study we show that the addition of glucose to the medium improves the growth of wild-type cells at high pH, fully abolishes the snf1 alkali-sensitive phenotype and attenuates high pH-induced Snf1 phosphorylation at Thr210. Lack of Elm1, one of the three upstream Snf1 kinases (Tos3, Elm1 and Sak1), markedly increases alkali sensitivity, whereas the phenotype of the triple mutant tos3 elm1 sak1 is even more pronounced than that of snf1 cells and is poorly rescued by glucose supplementation. DNA microarray analysis reveals that about 75% of the genes induced in the short term by high pH are also induced by glucose scarcity. Snf1 mediates, in full or in part, the activation of a significant subset (38%) of short-term alkali-induced genes, including those encoding high-affinity hexose transporters and phosphorylating enzymes. The induction of genes encoding enzymes involved in glycogen, but not trehalose, metabolism is largely dependent of the presence of Snf1. Therefore the function of Snf1 in adaptation to glucose scarcity appears crucial for alkaline pH tolerance. Incorporation of micromolar amounts of iron and copper to a glucose-supplemented medium resulted in an additive effect and allows near-normal growth at high pH, thus indicating that these three nutrients are key limiting factors for growth in an alkaline environment.
- gene expression
- glucose starvation
- stress tolerance
Fermentation of sugars, preferably glucose, is the favoured energy source for budding yeast. When glucose becomes limiting, Saccharomyces cerevisiae must adapt its metabolism accordingly. For instance, hundreds of genes modify their expression profile in a matter of minutes when S. cerevisiae is switched from glycerol to glucose as carbon source (i.e activation of genes encoding ribosomal proteins and inhibition of genes required for oxidative phosphorylation) . Depletion of glucose from the medium leads to the inverse effect .
Adaptation to different carbon sources requires the function of at least three major signalling pathways: the Ras/PKA (protein kinase A), the Snf1 and the Rgt networks. Snf1 is a protein kinase similar to mammalian AMP-activated protein kinase that is required for adaptation to glucose limitation and for utilization of alternate, less preferred, carbon sources. In response to glucose limitation, Snf1 is phosphorylated at Thr210 by three upstream Snf1 kinases (Sak1, Elm1 and Tos3), leading to Snf1 activation (for a review see ). Snf1 activity is negatively regulated by the type 1 protein phosphatase Glc7–Reg1 complex, which dephosphorylates Thr210 under conditions of ample glucose supply [4,5]. In cells lacking Reg1, the Snf1 catalytic activity is not inhibited by the presence of glucose in the medium . Recently, the type 2A-like phosphatases have also been implicated in Snf1 dephosphorylation [7,8]. Activation of Snf1 allows expression of many glucose-repressed genes, such as high-affinity glucose transporters (HXT2 and HXT4), and genes required for gluconeogenesis, as well as utilization of ethanol, lactate and other alternative carbon sources (i.e. SUC2). This is achieved by activation of positive transcription factors, such as Adr1, Cat8 or Sip4, inhibition of transcriptional repressors, such as Mig1, or stimulation of chromatin remodelling (for recent reviews see [3,9]). In essence, Snf1 represents a key player in the process of adaptation to glucose shortage. Consequently, a Snf1-deficient strain cannot grow on low levels of glucose (0.05%) or non-fermentable carbon sources. In conjunction with the Snf1 network, the Rgt1 network couples expression of the hexose transporters with glucose availability (for review see [10,11]). Glucose binding to two membrane sensors, Snf3 and Rgt2, induces their binding with the co-repressors Mth1 and Std1 and, hence, their recruitment to the membrane, where they are phosphorylated by the Yck1 and Yck2 casein kinases. Phosphorylation targets the co-repressors, by means of the SCFGrr1 ubiquitin conjugating complex, to degradation by the proteasome. Destruction of the co-repressors allows phosphorylation of Rgt1 and their eviction from the target promoters, thus releasing its repressive effect.
In addition to its relevance for carbon stress, Snf1 has been related to other type of stresses. For instance, an snf1 mutant shows defects upon phosphate starvation and this kinase is activated by nitrogen limitation [12,13]. Lack of Snf1 also confers sensitivity to genotoxic stress , Na+ and Li+, as well as other toxic cations, such as hygromycin B [15–19] or alkaline pH stress [18,19]. It has been proposed that the mechanisms by which Snf1 controls acquisition of sodium tolerance are different from those mediating glucose regulation . Exposure to diverse type of stresses (including Na+ and high pH) results in increased Snf1 phosphorylation and activation . However, whereas Snf1 rapidly translocates to the nucleus after 5 min of alkalinization of the medium, it does not do so upon Na+ stress .
It has been demonstrated that adaptive response of S. cerevisiae to high pH stress involves extensive gene remodelling, as a result of the activation of a number of signalling pathways, including the Rim101/Nrg1, the Wsc1/Pkc1/Slt2 MAPK (mitogen-activated protein kinase) and the calcium-activated calcineurin/Crz1 (calcineurin-responsive zinc finger 1) pathways [15,18,21–25]. Alkalinization also affects iron, copper and phosphate homoeostasis [21,22,26]. It has been proposed that the availability of copper and iron is a major limiting factor for proliferation of S. cerevisiae in alkaline growth conditions , as documented by the fact that addition of micromolar concentrations of these cations to the medium significantly improves growth in such environments. Previous work in our laboratory [24,25] pointed out that a large number of genes involved in hexose transport and carbohydrate metabolism were induced after short exposure of the cells (10–15 min) to alkaline pH. Many of these genes were also induced when cells were shifted to low-glucose-containing medium or at the diauxic shift , thus suggesting a link between alkaline stress and glucose utilization. Recent work in our laboratory has explored the role of PKA in high pH response and shown that it is largely based on Msn2/4 activation . On the other hand, we also demonstrated that a significant fraction of the glucose-related genes induced by alkaline stress in the short term were under the direct positive regulation of the calcineurin pathway, through activation of its downstream transcription factor Crz1 , and that activation of calcineurin was sufficient to allow growth of an snf1 mutant in low-glucose (0.05%) medium. These findings prompted us to propose that alkalinization of the medium could prevent proper glucose utilization and that activation of Snf1 and calcineurin would represent combined but independent strategies to deal with this problem. In the present paper, we test these hypotheses by investigating the contribution of the Snf1 kinase to the response to high pH. Our results suggest that Snf1 is responsible for a significant fraction of genes induced by high pH and that the alkaline growth defect of snf1 cells is essentially related to the role of this kinase in the control of glucose metabolism.
MATERIALS AND METHODS
Escherichia coli and yeast growth conditions
E. coli DH5α cells were used as plasmid DNA host and were grown at 37°C in LB (Luria–Bertani) medium with 50 μg/ml ampicillin when needed for plasmid selection. Yeast cells were grown at 28°C in YP medium (10 g/l yeast extract and 20 g/l peptone) containing in each case the specified amount of glucose or, when indicated, in synthetic minimal or complete minimal media . Growth tests were performed on plates (dot tests) essentially as described previously , including 50 mM Taps adjusted at the desired pH with KOH. In the case of the Snf1-activating kinases, dilutions of the cultures were made in a 50 mM sodium citrate (pH 3.0) and 5 mM EDTA solution, to minimize possible artefacts derived from the tendency of elm1 mutants to sediment. These cultures were also corrected for the number of cells by colony counting. Growth tests in liquid cultures were performed essentially as described previously .
Gene disruptions and plasmid construction
The yeast strains used in the present study are described in Table 1. Deletion mutants in the BY4741 background were generated by the Saccharomyces Genome Deletion Project . Strains with single tos3, elm1 and sak1 mutations, or their combination, were constructed in the W303-1A background and provided by Dr Enric Herrero (Universitat de Lleida, Lleida, Spain). Strain YL1157 was a gift from Professor Marian Carlson (Columbia University, New York, NY, U.S.A.). Plasmids YIp1242 , containing the entire HXK2 promoter cloned in YIp356, and YIp944 , containing the GLK1 promoter in the YIp358R plasmid, were a gift from Dr Pilar Herrero and Dr Fernando Moreno (Universidad de Oviedo, Oviedo, Spain). Plasmid pLS11 containing the entire SUC2 promoter was described previously . The reporter plasmid pHXK1-lacZ was generated as described in . The HXT2-LacZ reporter was a gift from Dr Sabire Ozcan (University of Kentucky, Lexington, KY, U.S.A.) and is described in [28,35]. Construction of the pPHO84-LacZ plasmid was reported previously .
Glucose uptake determination
Glucose uptake was evaluated as follows. Wild-type BY4741 cells (80 ml) were grown on YPD medium [1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose] up to A660=0.7 and centrifuged for 5 min at 1620 g. The cells were resuspended in 0.5 ml of YPD medium (pH 5.5) and then split into two aliquots of 0.25 ml. To the first aliquot (non-induced) 11 μl of 0.5 M KCl plus 25 μl of a [U-14C]glucose (0.2 mCi/ml, 311 mCi/mmol) solution was added, whereas the second aliquot (induced) received 11 μl of 0.5 M KOH (thus raising the pH to 8.0) plus the same amount of radioactive glucose. Samples (45 μl) were taken at given times, and immediately filtered through GF/C glass microfibre filters (25 mm diameter; Whatman) with the addition of 10 ml of ice-cold 0.5 M glucose solution. Filters were washed again with 10 ml of the glucose solution and introduced into scintillation vials for counting.
Samples for glucose-6-phosphate and fructose-6-phosphate measurements were taken after 10 min of exposure to pH 8.0 and determination of metabolites was carried out essentially as described in .
β-Galactosidase reporter assays
Yeast cells were grown to saturation in the appropriate dropout media and then inoculated into YPD 4% glucose (YP medium plus 4% glucose). Growth was resumed until A660=0.5–0.7, and the cultures were centrifuged for 5 min at 1620 g. Cells were resuspended in YPD 4% glucose (no induction), YPD 4% glucose plus 50 mM Taps adjusted to pH 8.0 (alkaline stress) or YPD 0.05% glucose (low glucose) and growth was resumed for 60 min. In all cases, β-galactosidase activity was measured as described previously .
For RNA preparation, yeast cells (DBY746 background) were grown on YPD medium (pH 5.5) to an absorbance (A660) of 0.5–0.8 and split into aliquots. Each aliquot was incubated 10 min either in YPD medium plus 20 mM KCl (non-stressed cells) or YPD medium containing 20 mM KOH (stressed cells, approximately pH 8.0). Measurement of cell viability by colony counting demonstrated that this treatment does not alter at all the viability of the culture. For glucose starvation experiments aliquots were centrifuged (1620 g for 5 min at 22°C) and cells were washed twice and resuspended in either YP medium containing 2% (non-induced) or 0.05% (starved) glucose. In all cases cultures were rapidly filtered through 0.45 μm Metricel membrane filters (Pall) and dried cells were kept at −80°C. Total RNA was extracted by using the RiboPure™ Yeast kit (Ambion) and treated with DNase to eliminate genomic DNA traces. RNA quality was assessed by denaturing 0.8% agarose gel electrophoresis, and RNA quantification was performed by measuring A260 in a BioPhotometer (Eppendorf).
DNA microarray analysis and RT (reverse transcription)–PCR
Total RNA (8 μg) was employed for cDNA synthesis and labelling, using the indirect labelling kit (CyScribe Post-Labeling kit; GE Healthcare), in conjunction with Cy3 (indocarbocyanine)–dUTP and Cy5 (indodicarbocyanine)–dUTP fluorescent nucleotides. The cDNA obtained was dried and resuspended in hybridization buffer. DNA amount and labelling was evaluated with a Nanodrop spectrophotometer (Nanodrop Technologies). Fluorescently labelled cDNAs were combined and hybridized to yeast genomic microchips constructed in our laboratory by arraying 6014 different PCR-amplified open reading frames from S. cerevisiae [24,38]. Prehybridization, hybridization and washing conditions were essentially as described previously . Slides were scanned with a ScanArray 4000 apparatus (Packard BioChips Technologies) and the output analysed using GenePix Pro 6.0 software. Several types of experiments were performed: (i) comparison of expression profiles of cells treated with KCl for 10 min with those of KOH-treated cells (pH 8.0) in wild-type cells (genetic background BY4741); and (ii) the same experiment using the snf1 mutant derivative. In each case, two independent biological replicates were performed, each by duplicate (dyes were swapped to avoid dye-specific bias). For each experiment, data were combined and the mean was calculated; (ii) comparison of expression profiles for wild-type cells transferred for 15 min from 2% to 0.05% glucose; and (iv) the same experiment for snf1 cells. For the glucose starvation experiments four microarrays were used. A given gene was considered to be induced or repressed when the expression ratio was higher than 2.0 or lower than 0.5 respectively. According to the expression of these genes in the snf1 strain, different levels of Snf1 dependence were established. Thus genes showing a ratio 0.67>X>0.50 were considered WD (weakly dependent), those with a ratio 0.50>X>0.25 were ranked as SD (strongly dependent) and those with a ratio ≤0.25 were labelled as TD (totally dependent). Similarly, genes induced more than 2.5-fold in wild-type cells and considered not induced (i.e. the ratio of high pH/low pH was <1.3) in snf1 cells were also considered as totally dependent. Microarray data can be retrieved from GEO (Gene Expression Omnibus) under the accession code GSE25697.
Changes in the expression of several genes were confirmed by RT–PCR analyses using the Ready-To-Go RT–PCR Beads kit (GE Healthcare) and 0.2 μg of total RNA. Gene-specific pairs of oligonucleotides (Supplementary Table S1 at http://www.BiochemJ.org/bj/444/bj4440039add.htm) were used to determine, after 25–30 cycles, the levels of the corresponding mRNAs.
Analysis of the possible transcription factor dependence of Snf1-regulated genes was carried out by crossing our microarray data with the YEASTRAC database . Only genes for which evidence for regulation by the transcription factor was both documented and direct were taken into account. Data were processed using Microsoft Access.
Immunodetection of phosphorylated Snf1
Cells (strain YL1157) were grown to exponential phase (A600=0.8) in YP medium containing 2%, 4%, 6% or 8% glucose, and an aliquot of each culture was collected by rapid filtration and frozen in liquid nitrogen. Another aliquot was collected by rapid filtration and resuspended in YP medium plus 2%, 4%, 6% or 8% glucose respectively, containing 100 mM Hepes (pH 8.0), for 10 min and collected by filtration and frozen. Whole-cell extracts were prepared as described previously , and proteins were separated by SDS/PAGE (8% gel) and analysed by immunoblotting using anti-[phospho-AMPK (AMP-activated protein kinase) (Thr172)] antibody (Cell Signaling Technology) to detected phosphorylation of Snf1 at Thr210. Membranes were incubated in 0.2 M glycine (pH 2), for 10 min and reprobed with anti-HA (haemagglutinin) antibody (12CA5, Roche) to detect Snf1–HA. ECL (enhanced chemiluminescence) Plus (GE Healthcare) was used for visualization.
High pH sensitivity caused by lack of SNF1 is eliminated by increasing glucose in the medium
Exposure to alkaline pH results in increased expression of many genes that are also responsive to glucose scarcity, thus suggesting that this situation activates pathways that are important in adaptation to lack of glucose. To evaluate the possible involvement of the Snf1 kinase, we tested the sensitivity to high pH of a number of mutants in genes related to this signalling pathway. As shown in Figure 1(A), mutation of SNF1 results in a patent increase in sensitivity. Similarly, lack of the plasma membrane glucose sensor SNF3, and of GRR1, a downstream component of the pathway required for regulatory glucose response, also results in enhanced high pH sensitivity. Interestingly, mutation of RGT2, encoding a glucose receptor highly similar to Snf3, only results in a weak sensitive phenotype. These results indicated that integrity of both Snf1 and Snf3/Rgt2–Grr1-mediated pathways was required for normal tolerance to high pH. Because these pathways are necessary for adaptation to low-glucose conditions, we considered the possibility that the observed sensitivity might reflect a situation of insufficient glucose availability. To test this possibility, we first considered if increasing the amount of glucose in the medium would affect tolerance to high pH. As shown in Figure 1(B), an increase in the concentration of the sugar in the medium in the range 1–5% positively affects growth when the medium is buffered to pH 8.0, whereas it has no effect in cultures grown at standard pH (5.5). In any case, even at the highest concentration tested (8%), glucose could not fully counteract the effect of high pH on growth rate, indicating the existence of additional limiting factors. We also investigated the possibility that shifting cells to high pH could compromise glucose uptake. To this end, glucose incorporation was evaluated by adding uniformly labelled [14C]glucose in YPD medium-grown cells at the same time that the pH of the medium was raised to 8.0 by addition of KOH. Figure 1(C) shows that glucose uptake is not negatively affected during the first few minutes of shifting to alkaline pH and, in fact, uptake appears even more effective after 6 min in cells subjected to stress. Measurement of the upper glycolytic intermediates glucose 6-phosphate and fructose 6-phosphate after 10 min of stress indicated that the concentration of these metabolites was not decreased in alkali-treated cells (results not shown).
We then compared growth of wild-type cells and the snf1 mutant at high pH (8.0) in the presence of increasing concentrations of glucose in the medium. As shown in Figure 2(A), the snf1 mutant is extremely sensitive to high pH at lower-than-normal glucose levels (1%, half the standard concentration in YPD medium), but this sensitivity is relieved by increasing glucose in the medium, reaching wild-type growth when the starting glucose concentration is 4%. It is known that high pH stress results in rapid phosphorylation of Snf1 at Thr210 that leads to activation of the kinase. Therefore we investigated if this signalling input could be influenced by the presence of additional glucose in the medium. As observed in Figure 2(B), doubling the amount of glucose in the medium substantially decreases the phosphorylation level of Snf1, suggesting that the stressing input has decreased. Interestingly, further increase in the amount of glucose has no additional effect on Snf1 phosphorylation state. These results indicate that the alkali-sensitive phenotype of the snf1 mutant is fully derived from the inability of this strain to cope with the alteration in normal glucose metabolism caused by increasing pH in the medium.
Evaluation of the role of upstream Snf1 protein kinases in high pH tolerance
Because Snf1 can be phosphorylated and activated by three different protein kinases (Sak1, Elm1 and Tos3) we investigated the contribution of these enzymes in high pH stress tolerance. As shown in Figure 3, Tos3 and Sak1 seem to have little or no contribution, since its mutation does not alter alkaline pH tolerance. In contrast, deletion of ELM1 yields a marked phenotype, albeit not as strong as that observed in the snf1 mutant. The sak1 tos3 double mutant shows no phenotype, whereas deletion of TOS3 and, particularly, SAK1 in the elm1 background further increases high pH sensitivity. Remarkably, the triple mutant is even more sensitive than the snf1 strain. Increasing the concentration of glucose from 2% to 6% markedly improved growth of most kinase mutants, but was largely ineffective in rescuing the strong phenotype of the strain lacking the three Snf1 upstream kinases.
The role of Snf1 in alkaline-pH triggered transcriptional response
As mentioned above, alkalinization of the medium results in transcriptional changes that affect many genes related to carbohydrate metabolism. We considered that if signalling pathways responding to glucose scarcity were contributing to the high pH response, then the kinetics of the response of these genes to high pH and to limiting glucose in the medium should be similar. To test this hypothesis, the expression of the lacZ gene from the HXT2 promoter was monitored against time after shifting cells to high pH or 0.05% glucose. We observed (Figure 4) that the initial rate of production of β-galactosidase activity was identical for cells exposed to high pH or subjected to low glucose. The decay in β-galactosidase production at longer times in cells stressed by high pH can be explained by a decrease in the pH of the medium along the time (reaching ~pH 7.4 after 120 min), which minimizes the stress condition (a situation that does not occur in cells exposed to low glucose). The profile of PHO84, a late-response high pH-inducible gene that does not respond to glucose starvation, is shown for comparison.
The transcriptional response to high pH is a relatively fast process, with a peak around 10–15 min [22,24], as it occurs with the activation of the Snf1 protein kinase. Therefore it was reasonable to evaluate the contribution of Snf1 to the short-term transcriptional response to alkaline pH stress. To this end, total RNA was prepared from wild-type and snf1 strains exposed for 10 min to pH 8.0, and the global expression level was compared with that of non-stressed cells using DNA microarrays. From 5259 genes with valid data for wild-type and snf1 cells, exposure of wild-type cells to high pH resulted in increased expression of 391 genes above the 2.0-fold threshold (Table 2 and Supplementary Table S2 at http://www.BiochemJ.org/bj/444/bj4440039add.htm). As shown in Table 2, 152 genes (38.9%) displayed some degree of Snf1 dependence in their activation upon high pH stress. For 21 of the genes the presence of Snf1 was an absolute requirement for induction (TD), whereas expression of 43 other genes was greatly reduced in the absence of the kinase (SD). The induction of 236 genes (61.1%) was essentially independent of Snf1, and three genes (0.8%) were actually more induced by high pH in the snf1 mutant than in the wild-type strain (with a snf1/wild-type ratio of induction >1.5). The entire set of induced genes can be found in Supplementary Table S2.
Gene ontology analysis of genes induced by high pH in the short-term (Table 3) indicates a single superfamily of genes highly represented: those involved in carbohydrate metabolism (P=5.95×10−10, not including sugar transporters). Among the carbohydrate-related genes, we find those encoding enzymes of trehalose metabolism (mostly biosynthesis, P=1.04×10−5), glycogen biosynthesis (P=3.81×10−5) as well as sugar transport and phosphorylation. Our data indicate that, whereas the expression of genes involved in glycogen metabolism is substantially influenced by the absence of Snf1, those related with trehalose metabolism are largely independent of Snf1 (Table 3 and Figure 5A). Other genes related to energy generation [TCA (tricarboxylic acid) cycle, mitochondrial, etc.)] also appear to be induced, including CAT8 and ADR1, which encode transcription factors important for adaptation to diauxic shift.
Exposure to high pH resulted in lower mRNA levels for 341 genes (Supplementary Table S3 at http://www.BiochemJ.org/bj/444/bj4440039add.htm). Gene ontology analysis shows a dramatic decrease in expression for genes involved in ribosome biogenesis (P=7.33×10−93), including a large number of those encoding ribosomal proteins (P=5.31×10−33) and rRNA-processing proteins (P=4.48×10−44). In the majority of the cases, high pH-induced repression was independent of Snf1 (252 genes, 73.9%). However, we identified 86 genes repressed by high pH in the wild-type, but not in the snf1 mutant (5 TD, 20 SD and 61 WD). Remarkably, whereas genes encoding rRNA processing proteins were found among both Snf1-dependent and -independent subsets, regulation of genes encoding ribosomal proteins turned out to be almost without exception independent of the presence of the kinase (P=2.40×10−40 for the Snf1-independent set and P=9.58×10−1 for the Snf1-dependent one).
We considered that if high pH stress is rapidly mimicking a situation of glucose starvation it should be possible to identify a significant overlap between the set of genes induced under both short-term alkaline stress and short-term glucose starvation. To this end, wild-type and snf1 cells were shifted to 0.05% glucose for 15 min and the expression profile was analysed by DNA microarray. A total number of 5131 genes were found to have valid data for wild-type cells subjected to high pH stress or low glucose. Among them, 390 were induced at least 2-fold by high pH and 797 by shifting to low glucose. When both subsets were compared, 293 common genes were found, a value 4.8-fold higher than that expected if both responses were unrelated. When this gene list was subjected to gene ontology analysis, an excess of genes related to carbohydrate metabolism (glucose metabolic process, P=9.04×10−9) was found, suggesting again a common biological alteration. In fact, most genes shown in Table 2 as induced by pH were also induced after a 15 min shift into low glucose. When the subset of snf1 cells was incorporated to the analysis a total number of 3627 genes with valid data was obtained (Figure 5B). Among them, 240 genes were found to be induced by both high pH and low glucose. A further 95 genes (39.6%) showed some degree of Snf1 dependence for high pH-triggered induction, whereas Snf1 dependence rose up to 65.4% for shift to low glucose (157 genes). Induction of 69 genes was Snf1 dependent for both high-pH- and low-glucose-stress. Therefore a large percentage (72.6%) of the genes whose induction by high pH was Snf1 dependent was also Snf1 dependent for low glucose. All of these data reinforce the notion that high pH stress triggers a situation that mimics a condition of glucose shortage and that the Snf1 kinase is likely to play an important role in gene activation upon alkaline pH stress.
We wished to confirm some of the results obtained from the DNA microarray analysis by alternative methods. To this end, LacZ translational reporter fusions were constructed for HXK1, GLK1 and SUC2, which, according to our microarray data, were induced by high pH in an Snf1-dependent form. An equivalent HXK2–LacZ reporter was tested as a control for a related, non-inducible gene. As shown in Figure 6(A), HXK1, GLK1 and SUC2 were potently induced by both high pH and low glucose, and this effect was largely abolished in the absence of Snf1. As expected, the activity of the HXK2 reporter was almost unaltered by exposure to high pH. The influence of the presence or absence of Snf1 on the alkaline pH-induced expression of other selected genes, such as MDH2, HXT7, GAL2, ALD4, MAL33 and again SUC2 was monitored by RT–PCR. The results, shown in Figure 6(B), confirm the diverse level of Snf1 dependence inferred from the microarray analysis.
Finally, we considered that it could be interesting to examine the possible relevance in the alkaline response of transcriptional activators or repressors that are known to be controlled by Snf1. To this end we scanned the YEASTRACT database for genes known to be controlled by Adr1, Mig1, Mig2, Nrg1 and Nrg2 and crossed these datasets with our list of induced genes. As shown in Table 4, only a small number high pH-induced genes are also Mig1- or Mig2-regulated genes. In contrast, the intersection for Nrg1, Nrg2 or Adr1 yields a higher number of common genes which is 1.6–2.4 in excess when compared with a random distribution. It is worth noting that, even in these cases, 40–50% of the genes controlled by the transcription factors are not dependent of Snf1 for alkaline induction. This could be explained if it is considered that Nrg1 and Nrg2 can be also under the control of the Rim101 pathway, which is activated by alkaline stress [18,21,23]. Similarly, Adr1 function could be negatively regulated by PKA (see ) and it has been shown recently that alkalinization of the medium induces fast inhibition of the PKA pathway .
The Rgt signalling network is also important for response to high pH stress
The phenotypic analysis shown in Figure 1 indicates that, in addition to the snf1 mutation, lack of the Snf3 glucose membrane sensor or Grr1, a downstream component of the pathway, results in increased sensitivity to alkaline pH. We considered it interesting to evaluate if the absence of these components would affect the transcriptional response of a gene relevant for adaptation to low-glucose conditions, such as the high-affinity hexose transporter HXT2, which is strongly induced by high pH. To this end, plasmid pHXT2-LacZ was introduced into the appropriate strains and the cultures subjected to high pH or low glucose stress. As shown in Figure 6(C), lack of Snf1 partially blocks the response to high pH of HXT2 (thus confirming the microarray data) and totally abolishes its induction by low glucose . Interestingly, lack of Snf3 also blocks to some extent high pH-induced expression from the HXT2 promoter. The absence of Grr1, which results in enhanced basal expression from the promoter, also interferes with full induction upon exposure to alkaline pH. Therefore integrity of the both the Snf1 and Rgt pathways seem important for normal tolerance and full response to high pH stress.
Additive effect of glucose and iron/copper supplementation on growth at high pH
Previous work in our laboratory demonstrated that copper and iron availability were limiting factors for yeast growth under alkaline pH stress, likely due to a decrease in the solubility of these metals, by showing that addition of micromolar amounts of both metals to the medium dramatically improves growth at high pH . We show in the present study that addition of glucose also improves growth under these conditions. Therefore two relevant questions arise: (i) are the beneficial effects of iron/copper and glucose addition independent events?; and (ii) are copper/iron and glucose availability the only limiting factors for growth at alkaline pH? To address these questions, wild type cells were grown at pH 5.5 and 8.0 in the presence of different concentrations of glucose and in the presence or absence of a mixture of 5 μM (NH4)2Fe(SO4)2 and 5 μM CuSO4. As shown in Figure 7, increasing glucose concentration in the medium from 2% to 4% further increases the growth rate promoted by inclusion of copper and iron, indicating that both effects are additive. Additional increase of glucose in the medium to 6% resulted in quantitatively identical results (results not shown). The improvement on growth rate caused by supplementing the medium with copper/iron plus higher glucose amounts was rather important, although it did not match growth at acidic pH.
Previous evidence suggested that the response to high pH stress involves the Snf1 kinase pathway [15,18,19,28], and fast activation of the kinase after alkalinization of the medium was demonstrated . Work in our laboratory [24,25,28] indicated that a significant subset of changes in the transcriptional profile in response to high pH were compatible with alterations in the utilization of glucose and in carbohydrate metabolism, where Snf1 plays a very relevant role. In the present study we characterize the contribution of the Snf1 kinase to the adaptive high pH response. Evaluation of the snf3 and grr1 phenotypes suggests that impairing glucose signalling upstream Snf1 decreases high pH tolerance and further reinforces the notion that correct glucose sensing is important for normal high pH tolerance. The observations that increasing the starting amount of glucose in the medium to 4% attenuates the high-pH-induced phosphorylation of Snf1 and fully relieves the sensitivity of this mutant to alkaline pH provides further support to the concept that the incapacity to deal with the alteration in glucose utilization caused by increasing pH in the medium is at the origin of the high pH-sensitive phenotype of the snf1 mutant.
We also explored the relative contribution of the upstream Snf1 phosphorylating kinases. The observation that the strain lacking all three kinases is strongly sensitive to high pH confirms the findings of Hong and Carlson  and is in agreement with those of Ye, Elbing and Hohmann  regarding tolerance to NaCl. However, in our case we observe a rather strong high pH phenotype for the elm1 strain, which had not been reported previously. Remarkably the alkaline pH-sensitive phenotype for the sak1 elm1 tos3 strain is stronger than that of the snf1 strain and is not completely rescued by addition of extra glucose. The lack of rescue is also observed in the combination of the elm1 mutation with that of TOS3 or SAK1 genes (Figure 3), suggesting that the Elm1 kinase could have an important function in high pH tolerance besides its regulatory role on the Snf1 kinase. It is worth considering that Elm1 has been shown to be important for normal cellular morphogenesis . Because alkalinization of the medium imposes a stress on the cell wall, activating the Slt2 MAPK pathway , it is plausible that the strong phenotype of the elm1 cells could be the result of the combination of deficient Snf1 signalling and morphogenetic defects aggravated by cell-wall stress.
Previous work in our laboratory [24,25] and the results of the present study indicate that the overall profile of the short-term response to high pH mimics that observed upon glucose starvation, with marked increase in the expression of high-affinity glucose transporters and TCA-cycle-related genes, suggesting a shift to respiratory metabolism. Our data could provide the basis to understand a previous report  in which long-term carbon fluxes under specific perturbations (including alkalinization of the medium) were examined. These authors observed that yeast cells growing in batch, when shifted to pH 7.5, accumulated acetate and showed an increase in activity of the TCA cycle and respiration rate. This behaviour fits well with the potent induction of ALD4 (~6.5-fold), combined with more modest changes in ALD5 and ALD6 (~2-fold), which encode aldehyde dehydrogenases that produce acetate from acetaldehyde, as well as with the increased expression of ACS1, encoding acetyl CoA synthetase, the enzyme that converts acetate into acetyl CoA. We also observe the induction of PYC1, encoding pyruvate carboxylase, which transforms pyruvate into oxalacetate, an intermediate of the TCA cycle. The last two activities are important for supplying carbons to this cycle and, together with the induction of genes encoding several enzymes belonging or related to the cycle itself, such as CIT2 (>10-fold), CIT1, LSC2, MDH1 or MDH2, may explain the observed changes in metabolic fluxes. It must be noted, however, that although the metabolic changes observed in  upon shifting cells to pH 7.5 were coincident with a decrease in glucose uptake rate, this was not observed under our conditions (Figure 1C). Therefore the short-term transcriptional response described in the present paper cannot be attributed to a deficient entry of glucose and could be the result of the incapacity of the cell to properly metabolize the sugar under alkaline conditions. In any case, our data suggest that this hypothetical alteration is quite effectively overridden by increasing the amount of available glucose in the medium.
Our DNA microarray data strongly support the participation of Snf1 in this short-term transcriptional response, in agreement with the observation that Snf1 becomes enriched in the nucleus shortly after glucose-grown cells are shifted to non-fermentable carbon sources or to alkaline pH . It is worth noting the differential regulatory network controlling expression of genes involved in trehalose and glycogen metabolisms in response to high pH. Genes involved in trehalose metabolism are induced by high pH predominantly in an Snf1-independent fashion (Table 3 and Figure 5A) and recent work from our laboratory has shown that this change in expression is largely Msn2/Msn4 dependent . In the present study we show that induction by alkaline pH of multiple genes related to glycogen metabolism is dependent on both the presence of Snf1 and Msn2/Msn4 (Table 3 and Figure 5A). The impact of the Snf1 and PKA–Msn2/Msn4 pathways has been studied in the case of expression of glycogen synthase, encoded by the genes GSY1 and GSY2 (see  for a recent review). Both genes integrate inputs from Snf1 and PKA in response to glucose availability, although the precise target for Snf1 in GSY2, encoding the major glycogen synthase isoform, is not known. Our data indicate that, in response to high pH stress, this dual signalling also influences many genes encoding glycogen-metabolizing enzymes. This regulation affects not only genes encoding activities directly involved in synthesis or degradation of glycogen, such as GLC3 and GDB1, encoding the branching and debranching activities respectively, or GLG1, encoding one of the two glycogenins, but also components of the post-translational regulatory network, such as GIP2 or GAC1. These genes encode related regulatory subunits of Glc7 [45,46], the major protein phosphatase activity acting on glycogen synthase . It is worth noting that UGP1, encoding the activity that synthesizes UDP-glucose, is induced by high pH in an Msn2/Msn4-dependent, but not in a Snf1-dependent, fashion. This cannot be considered unexpected, because UDP-glucose is also utilized as precursor for many other purposes in the cell, such as trehalose synthesis, cell-wall generation (synthesis of β-1,3 and β-1,3 glucans) or even protein N-glycosylation.
In conclusion, the present study collects a number of evidences indicating that high pH stress compromises correct utilization of glucose and that the role of Snf1 in tolerance to alkalinization is largely due to the function of the kinase in the adaptation of the cell to glucose scarcity, thus delineating with both the PKA  and calcineurin  pathways, the key elements for this specific response. In addition, the fact that the supplementation of the medium with glucose is additive to our previously described beneficial effect caused by micromolar concentrations of a mixture of copper and iron salts  indicates that these nutrients positively influence growth under alkaline conditions through different mechanisms. The growth rate at pH 8.0 obtained by addition of glucose plus copper/iron salts is still somewhat lower than that of cells inoculated at low pH, thus suggesting that additional limiting factors must exist. However, the improvement observed is rather impressive and it demonstrates that investigation of the signalling pathways that are activated in response to high pH stress can provide important clues for enhanced cell growth and culture yields under such adverse conditions.
Antonio Casamayor performed DNA microarrays experiments and analysis, and RT–PCRs. Raquel Serrano and Carlos Casado performed glucose uptake experiments. María Platara and Raquel Serrano contributed to LacZ reporter analysis. Carlos Casado carried out the phenotypic analysis of kinases acting on Snf1. Amparo Ruiz was responsible for phenotypic experiments, LacZ reporter analysis and Snf1 phosphorylation experiments. Joaquín Ariño conceived the project, supervised its development and wrote the paper.
This work was supported by the Ministry of Science and Innovation, Spain and FEDER [grant number BFU2008-04188-C03-01, BFU2011-30197-C3-01 and EUI2009-04147 (SysMo2) (to J.A.) and BFU2009-11593 (to A.C.)]. J.A. was recipient of an ‘Ajut 2009SGR-1091’ (Generalitat de Catalunya) and an ICREA Academia 2009 Award (Generalitat de Catalunya). C.C. was supported by a predoctoral fellowship from the Spanish Ministry of Science and Technology.
We thank Dr Enric Herrero (Universitat de Lleida, Lleida, Spain), Dr Pilar Herrero and Dr Fernando Moreno (Universidad de Oviedo, Oviedo, Spain), Dr Sabire Ozcan (University of Kentucky, Lexington, KY, U.S.A.) and Professor Marian Carlson (Colombia University, New York, NY, U.S.A.) for strains and reagents. We thank Laia Viladevall for early metabolite determinations. The support of A. González and M. Robledo is acknowledged.
Abbreviations: AMPK, AMP-activated protein kinase; Crz1, calcineurin-responsive zinc finger 1; HA, haemagglutinin; MAPK, mitogen-activated protein kinase; PKA, protein kinase A; RT, reverse transcription; SD, strongly dependent; TCA, tricarboxylic acid; TD, totally dependent; WD, weakly dependent
- © The Authors Journal compilation © 2012 Biochemical Society